Sum-Frequency-Generation Vibration Spectroscopy and Density

ACS2GO © 2018. ← → → ←. loading. To add this web app to the home screen open the browser option menu and tap on Add to homescreen...
0 downloads 0 Views 708KB Size
Article pubs.acs.org/JPCB

Sum-Frequency-Generation Vibration Spectroscopy and Density Functional Theory Calculations with Dispersion Corrections (DFT-D2) for Cellulose Iα and Iβ Christopher M. Lee,† Naseer M. A. Mohamed,‡ Heath D. Watts,‡ James D. Kubicki,‡ and Seong H. Kim*,†,§ †

Department of Chemical Engineering and Materials Research Institute, and ‡Department of Geosciences, The Pennsylvania State University, University Park, Pennsylvania 16802, United States S Supporting Information *

ABSTRACT: Sum-frequency-generation (SFG) vibration spectroscopy selectively detects noncentrosymmetric vibrational modes in crystalline cellulose inside intact lignocellulose. However, SFG peak assignment in biomass samples is challenging due to the complexity of the SFG processes and the lack of reference SFG spectra from the two crystal forms synthesized in nature, cellulose Iα and Iβ. This paper compares SFG spectra of laterally aligned cellulose Iα and Iβ crystals with vibration frequencies calculated from density functional theory with dispersion corrections (DFT-D2). Two possible hydrogen-bond networks A and B (Nishiyama et al. Biomacromolecules 2008, 9, 3133) were investigated for both polymorphs. From DFT-D2 calculations the energetically favorable structures for cellulose Iα and Iβ had CH2OH groups in tg conformations and network A hydrogen bonding. The calculated frequencies of C−H stretch modes agreed reasonably well with the peak positions observed with SFG and were localized vibrations; thus, peak assignments to specific alkyl groups were proposed. DFT-D2 calculations underestimated the distances between hydrogen-bonded oxygen atoms compared to the experimentally determined values; therefore, the OH stretching calculated frequencies were ∼100 cm−1 lower than observed. The SFG peak assignments through comparison with DFT-D2 calculations will guide the SFG analysis of the crystalline cellulose structure in plant cell walls and lignocellulose biomass.



and are deposited into the cell wall.11 There are two crystal structures that are produced in nature: cellulose Iα, predominantly found in algae and bacterial biofilms and cellulose Iβ, predominantly found in land plants and tunicates.12 In cellulose microfibrils, the six-membered glucan unit has the arm-chair conformation and the hydroxyl (O2H, O3H) and exocyclic hydroxymethylene (C6H2OH) groups at the equatorial positions, and the methine (CH) functional groups on carbons 1, 2, 3, 4, and 5 at the axial positions (Figure 1). Three conformations of the C6H2OH groups are possible through the rotation of the C5−C6 bond, the tg conformation is found to be the dominant form in cellulose Iα and Iβ as shown in Figure 1.13,14 Cellulose chains are linear and assembled side by side into a sheet through networks of intrachain hydrogen bonds (O3H···O5; O2H···O6) and interchain hydrogen bonds (O6H···O2′; O2H···O6′; O6H···O3′) (Figure 2). There are two possible patterns of hydrogen bonding within a sheet which are called networks A and B.15 In network A (Figure 2a), the hydrogen of O2 is involved in intrachain hydrogen bond to O6 (O2H···O6) and the hydrogen of O6 is involved in interchain hydrogen bonds

INTRODUCTION Vibration spectroscopy of cellulose has been an important tool in the biological sciences and also sustainable engineering to understanding cellulose structure.1 Plants use much of the carbon and solar energy captured during photosynthesis to produce cell walls, which are critical for proper growth and development.2 Plant biomass is processed for use as lumber, textiles, papers, and additives in foods, paints, and films.3 Recently, global efforts are underway to convert plant cell walls (which are collectively called lignocellulosic biomass) to biofuels for transportation needs.4−6 Cellulose is the most abundant and important constituent in plant cell walls. Its crystal structure and complex network with other carbohydrate polymers are key factors determining the mechanical strength and degradability of plant cell walls.7,8 A nondestructive spectroscopic technique that can selectively detect crystalline cellulose imbedded in amorphous carbohydrate polymers is necessary to advance our understanding of the roles of crystalline cellulose in the chemical and physical properties of plant cell walls. Cellulose is a linear polymer chain of β-D-glucopyranose monomers connected by 1,4 glycosidic linkages. In plants, chains are synthesized by cellulose synthase complexes (CSCs) present at the plasma membrane.9,10 Shortly after polymerization, several chains crystallize into elementary microfibrils © 2013 American Chemical Society

Received: March 26, 2013 Revised: May 8, 2013 Published: May 29, 2013 6681

dx.doi.org/10.1021/jp402998s | J. Phys. Chem. B 2013, 117, 6681−6692

The Journal of Physical Chemistry B

Article

plant cell walls are smaller (∼2−5 nm) and interact with hemicellulose, pectin, and lignin components in the cell wall.11,18 Determining the native cellulose polymorphism,19,20 crystallinity,21 hydrogen bonding,22−24 and mechanical properties25 is commonly performed using IR and Raman vibration spectroscopy as well as X-ray diffraction (XRD) and solid-state nuclear magnetic resonance (NMR).26−30 These techniques are useful for structural analysis of isolated or purified cellulose; but they suffer from spectral interferences from other matrix polysaccharide components when applied to the structural study of cellulose within intact plant cell walls.31−33 Because crystalline cellulose is more resistant to hydrolysis than amorphous polysaccharides, cellulose can be isolated from plant cell walls. However, this process can alter the crystal structure of native cellulose. In order to study crystalline cellulose structure in situ without separation of cellulose from other matrix components, it is desirable to use a spectroscopic technique that is highly selective to cellulose only. Recently, sum-frequency-generation (SFG) vibration spectroscopy has been shown to selectively detect crystalline cellulose in plant cell walls and lignocellulosic biomass without interferences from amorphous components such as hemicellulose, pectin, and lignin.33,34 SFG is a nonlinear optical process in which visible (ωVIS) and infrared (ωIR) photons are combined into a single photon whose frequency is the sum of the two input frequencies, ωSFG = ωVIS + ωIR. The output SFG signal I(ωSFG) is expressed as35−37

Figure 1. Molecular model of the exocyclic C6H2OH group in tg (green), gt (blue), and gg (pink) conformations. In the pairwise representation (t = trans and g = gauche), the first letter is the torsion angle of O5−C5−C6−O6 and the second letter the angle of C4−C5− C6−O6. χ is the angle between the C5−O5 and C6−O6 bond axes (C = dark gray; O = red; H = light gray).

(O6H···O3′).15 In network B (Figure 2b), the hydrogen of O2 is involved in interchain hydrogen bond to O6 (O2H···O6′) and the hydrogen of O6 is involved in intrachain hydrogen bond to O2 (O6H···O2).15 These sheets are stacked through van der Waals interactions between the hydrophobic facets of the glucan units to form three-dimensional crystals of cellulose Iα and cellulose Iβ.13,14 Sheet stacking influences the covalent bond lengths as well as the torsion angles of the glycosidic bonds and the hydroxymethylene groups in cellulose Iα and Iβ.16 In the cellulose Iα triclinic unit cell, all sheets are identical and the chains in each sheet contain alternating units with slightly different CH2OH conformations (χ = 166° and 167°, Figure 3b).14 In the cellulose Iβ monoclinic unit cell, chains in each sheet are identical, but the CH2OH conformations of the glucose units in two adjacent sheets are slightly different (χ = 158° and 170°, Figure 3c).13 The structural models described in Figure 3 are constructed based on X-ray and neutron diffraction studies of aligned nanowhiskers (cylinder-shape nanocrystals) of cellulose Iα isolated from the green alga Glaucocystis nostochinearum and cellulose Iβ isolated from the tunicate Halocynthia roretzi.13,14 These species are unique in that they produce large crystals (∼20 nm) of a single allomorph. Cellulose crystals in cell walls of vascular plants contain varying ratios of Iα and Iβ forms and are thought to be more disordered than cellulose from Halocynthia and Glaucocystis.17 In addition, cellulose in vascular

I(ωSFG) ∝ |χeff (2) |2 I(ω VIS)I(ωIR )

(1)

where I(ωVIS) and I(ωIR) are the intensities of the visible and infrared laser beams and χeff(2) is the effective nonlinear susceptibility:35 χeff (2) =

N ∑α , β , γ ⟨Mαβ A y ⟩ ε0(ωIR − ωq − i Γ)

(2)

Here, N is the number density, ε0 is the dielectric constant of vacuum, ⟨MαβAy⟩ is the angle-average of the product of the Raman and infrared tensors, ωq is the frequency of a normal vibration mode, and Γ is the damping constant. In the SFG analysis with a tunable IR beam, signals are resonantly

Figure 2. Two possible hydrogen-bonding networks in cellulose Iα and Iβ crystals: (a) network A and (b) network B15 (C = dark gray; O = red; H = light gray). 6682

dx.doi.org/10.1021/jp402998s | J. Phys. Chem. B 2013, 117, 6681−6692

The Journal of Physical Chemistry B

Article

Figure 3. Two conformations of the exocyclic CH2OH groups in cellulose Iα and Iβ are shown in (a). The χ angles are marked in red and blue colors. Stacking of cellulose chains in (b) cellulose Iα and (c) cellulose Iβ. The dotted boxes are the triclinic unit cell containing a single chain in (b) and the monoclinic unit cell containing two chains in (c). The unit cell axes are marked following the notation from Sugiyama et al.85 The c-axis in (b) is reversed for clarity.

valuable information needed for peak assignment of the SFG spectra. There have been several simulation studies to calculate the vibration frequencies for cellulose to assist peak assignments. Earlier works have used a classical mechanics approach called normal-coordinate analysis to compare calculated vibration modes with the IR and Raman spectra of cellulose from the alga Valonia ventricosa.49,50 Recently, ab initio calculations have been used to evaluate the crystal structures of native cellulose15,51 and to predict the vibrational spectra using Hartree−Fock52 and density functional theory (DFT).53,54 Barsberg et al. used cellulose I single-chain models in DFT calculations to estimate peak positions and intensities. However, these results did not distinguish between Iα and Iβ because only the covalent intrachain interactions were considered. A more advanced method is DFT with dispersion corrections (DFT-D2) that includes an empirical correction for long-range interactions.55 DFT-D2 has been recently applied to cellulose Iα and Iβ to calculate crystal structures, intermolecular forces, interactions with water, and allomorph conversion.56,57 In this study, we report the SFG spectra of highly crystalline and laterally packed cellulose Iα and Iβ nanowhiskers prepared from Glaucocystis and Halocynthia, respectively, and compare the SFG peaks with the fundamental vibration modes calculated from DFT-D2.54 The crystal structures of cellulose Iα and Iβ were constructed from the experimentally determined atomic coordinates and optimized through energy minimization processes. The vibration energies obtained from DFT-D2 calculations were rescaled with the empirically known peak positions of specific localized vibration modes. The comparison of the DFT-D2 calculations with the SFG spectra made the peak assignments possible for the exocyclic methylene group in cellulose Iα and Iβ crystal and the hydrogen-bonding network in each allomorph. Being able to detect and identify the vibration modes of crystalline cellulose in plant cell walls and lignocellulosic biomass will greatly assist the structural study for the roles of crystalline cellulose in plant growth and mechanics as well as recalcitrance of biomass to degradation processes.

enhanced as the frequency of the input IR beam (ωIR) approaches ωq in the sample. Equation 2 also states that the vibration mode must be both IR- and Raman-active. The thirdorder tensor, χeff(2), dictates that the SFG process takes place only when the irradiated medium has no inversion symmetry, otherwise χeff(2) is zero.38 Only certain vibrational modes in crystalline cellulose within plant cell walls meet this noncentrosymmetry.39 In contrast, other amorphous matrix polymers (hemicellulose and lignin) in plant cell walls do not produce SFG signal even though they are composed of monomers which contain several chiral centers.33 Without net polar ordering or electronic resonance, the SFG signals from chiral centers are very weak using reflection geometry, which is chosen due to opaque plant cell wall samples.40 For these reasons, SFG can selectively detect cellulose in lignocellulose biomass. Peak assignment is challenging, however, because several factors affect the SFG signal generation from crystalline cellulose. The nonlinear susceptibilities of individual vibration modes depend on the polarizations of input and output photons. If a single crystal or molecular thin film of cellulose is available, one could conduct thorough polarization analyses to obtain the χeff(2) tensors of various vibration modes.41,42 Unfortunately, native cellulose chains crystallize into only a few nanometers wide microfibrils and often cellulose Iα or Iβ crystals are present together in polycrystalline samples18 or are aggregated into bundles.29,43 Moreover, cellulose is a birefringent material containing many chiral centers in the glucan monomeric unit.44 Thus, the polarization of the light can be altered while light travels through the sample. 45 Furthermore, SFG intensity analysis is complicated because the phase matching condition is strongly influenced by the spatial distribution of cellulose crystals in the sample over the SFG coherence length.46 Hieu et al. proposed the χeff(2) tensors for cellulose in a single cotton fiber.39 In this method, it was assumed that the cellulose crystalline axis coincides with the fiber axis in the laboratory coordinates. However, the cellulose microfibril orientation in cotton fibers is known to contain several layers, each layer with alternating spiral arrangements.47 Thus, using SFG to analyze plant cell walls is complicated because the native cellulose structure could be influenced by the growth process.11,48 In this situation, theoretical calculations of vibration modes and frequencies could provide



METHODS Cellulose Sources. Films of highly aligned cellulose Iα and Iβ nanowhiskers, prepared by H2SO4 treatment of Glaucocystis nostochinearum and the tunicate Halocynthia roretzi, were 6683

dx.doi.org/10.1021/jp402998s | J. Phys. Chem. B 2013, 117, 6681−6692

The Journal of Physical Chemistry B

Article

Figure 4. Two laser incidence plane geometries for aligned cellulose Iα and Iβ nanowhiskers. The nanowhisker alignment axis (black arrow) of the cellulose films was positioned (a) parallel (∥) and (b) perpendicular (⊥) to the laser incidence plane (gray). The polarizations of IR and visible incidence beams were in-plane (p) and out-of-plane (s), respectively, with respect to the laser incidence plane.

incident light was normal to the sample and spectra were taken by rotating the sample so that the nanowhisker alignment axis was aligned parallel and perpendicular to the incident beam polarization. All spectra were taken in the region 250−3700 cm−1 using 8 cm−1/step and averaged over 1000 scans. DFT-D2 Calculations. Periodic DFT calculations were performed with the Vienna Ab-initio Simulation Package (VASP).59−62 The initial crystal structures of cellulose Iα and Iβ were created using Materials Studio 5.5 (Accelrys Inc., San Diego, CA) based on the coordinates determined by Nishiyama et al.13,14 The simulation cell size contained C24O20H40 for cellulose Iα and C48O40H80 for cellulose Iβ. These corresponded to two unit cells for cellulose Iα and Iβ, respectively. Cartesian coordinate files generated with Materials Studio 5.5 were converted into VASP 5.2 input files via a Perl script written by A. V. Bandura (St. Petersburg State University). Manipulations of the hydroxymethylene group torsions to three different conformations (tg, gg, and gt) and hydrogen-bonding networks A and B were performed manually in Materials Studio 5.5. Networks A and B were based on two possible hydrogen atom positions centered about 6O suggested by Nishiyama and co-workers.15 Projector-augmented planewave pseudopotentials were used with the PBE gradient-corrected exchange correlation function for the 3-D periodic DFT calculations. The choice of electron density and atomic structure optimization parameters was based on previous work.56,57 An energy cutoff of 800 eV was used with an electronic energy convergence criterion of 1 × 10−7 eV. Atomic structures were relaxed until the energy gradient was less than 0.02 eV/Å at 2 × 2 × 2 k-point samplings. Atoms were first allowed to relax with the lattice parameters constrained to the experimental values, then the atoms and lattice parameters were allowed to relax to obtain reported structures, energies and spectroscopic properties. The dispersion-correction parameters were 40 Å for the cutoff distance56 and 0.75 for the scaling factor (s6) and 20 for the exponential coefficient (d) in the damping function.55 Frequency analyses were performed on the energyminimized structures as predicted using VASP. Second derivatives of the potential energy matrix with respect to atomic displacements were calculated using two finite-difference steps (NFREE=2) and atomic movements of 0.015 Å (POTIM=0.015). Vibrational modes were analyzed using the program wxDragon 1.8.0.63 Representative movies are shown in the Supporting Information.

generously provided by Yoshiharu Nishiyama, CERMAV− CNRS, France. The details of sample preparation and full characterization of these samples with X-ray and neutron diffraction were previously published.13−15 In these samples, cellulose chains were roughly aligned with the unit cell c-axis along one direction by shearing of an aqueous suspension of nanowhiskers during the drying process.58 The SEM images and 2D X-ray diffraction data of these samples are shown Figure S1 in the Supporting Information. Avicel PH-101 with ∼50 μm particle size microcrystalline cellulose (FMC Biopolymer, CAS-No. 9004-34-6) was pressed into a pellet. SFG Vibration Spectroscopy. The SFG setup (EKSPLA) used in our laboratory was described in previous studies.33,34 The 532 nm laser pulse was generated by frequency doubling of the 1064 nm output from a Nd:YAG laser. An optical parameter generator/amplifier (OPG/OPA) using β-BaB2O4 and AgGaS2 crystals was pumped with 532 and 1064 nm laser pulses and generated tunable IR pulses in the wavelength range of 2.3−10 μm with